Manual Evaluation

Manual Evaluation is the simplest of the three evaluation methods: a reviewer looks at the candidate's response and enters one overall score for the question — no criteria, levels, or AI involved. It's what a question falls back to when neither a Rubric nor AI Evaluation is configured.

Entering a Score

The reviewer types a single value, entered either as a percentage of the question's positive multiplier or as a raw point value — whichever entry mode the review screen is set to. Because success rate can itself be negative, the entry also accepts negative values, letting a manual review apply the same negative-marking behavior an auto-scored question would get from its negative multiplier.

When to Use It

Manual Evaluation suits low-volume review, quick one-off judgment calls, or questions where a rubric would be overkill. As review volume grows, or when several reviewers need to score the same kind of response the same way, Rubric Scoring gives more consistency than a freehand number.

If AI Evaluation is also enabled on the question, entering a manual score overrides the AI's result for that response — a human judgment always wins over the AI's. See Evaluation & Grading for how the methods interact.

Good to know